Datasets:
ArXiv:
License:
Dataset Viewer
The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.
Search-Gen-V-evallog Dataset
This repository contains the running logs of the experiments conducted in the paper AN EFFICIENT RUBRIC-BASED GENERATIVE VERIFIER FOR SEARCH-AUGMENTED LLMS.
Related links
- paper:
- code:
- model:
- datasets:
Result
Table 1. Results on the eval set
Verifier Model Rubric Precision Rubric Recall Rubric F1 Sample Precision Sample Recall Sample F1 Avg. F1 Qwen3-1.7B 0.41 0.49 0.34 0.48 0.40 0.32 0.33 Qwen2.5-3B 0.42 0.47 0.43 0.49 0.46 0.43 0.43 Qwen3-4B 0.56 0.62 0.57 0.61 0.58 0.58 0.58 Qwen3-8B 0.54 0.66 0.55 0.62 0.61 0.57 0.56 LLaMA-3.1-8B 0.45 0.54 0.42 0.34 0.41 0.32 0.37 Qwen3-30B-A3B 0.56 0.66 0.56 0.63 0.62 0.62 0.58 Qwen2.5-32B-Instruct 0.60 0.67 0.60 0.67 0.68 0.64 0.62 Search-Gen-V-1.7B (SFT) 0.63 0.62 0.62 0.66 0.66 0.66 0.64 Search-Gen-V-4B (SFT) 0.70 0.66 0.68 0.72 0.72 0.71 0.70 Search-Gen-V-4B (SFT+RL) 0.71 0.68 0.70 0.74 0.74 0.73 0.72 Qwen3-235B-A22B-Instruct-2507 0.72 0.73 0.73 0.76 0.76 0.76 0.74 Table 2. Accuracy comparison on verifying rubrics in longform answers from DeepResearch Bench
Verifier Model Precision Recall F1 Qwen3-4B 0.42 0.56 0.42 Search-Gen-V-4B 0.59 0.57 0.57 Qwen3-235B-A22B 0.57 0.67 0.61 Table 3. Results on the short-form workload, HotpotQA
Verifier Model Precision Recall F1 EM 0.84 0.80 0.82 Qwen3-4B 0.83 0.70 0.71 Search-Gen-V-4B 0.86 0.76 0.77 Qwen3-235B-A22B 0.87 0.78 0.80 EM + Qwen3-4B 0.94 0.92 0.93 EM + Search-Gen-V-4B 0.95 0.93 0.94 EM + Qwen3-235B-A22B 0.96 0.94 0.95
Citation
@article{ma2025searchgenv,
title={AN EFFICIENT RUBRIC-BASED GENERATIVE VERIFIER FOR SEARCH-AUGMENTED LLMS},
author={Ma, Linyue and Xu, Yilong and Long, Xiang and Zheng, Zhi},
journal={arXiv preprint arXiv:2510.14660},
year={2025},
url={https://arxiv.org/abs/2510.14660}
}
- Downloads last month
- 84